A number of computational techniques have been offered for estimation of unmeasured states in nonlinear systems. Most of these techniques rely on applying the linear estimation techniques to the linearized systems, which can be effective only in the neighborhood of the operating point. This paper presents an online technique for nonlinear stochastic tracking problems. The idea of the proposed technique is to integrate the Kalman filter algorithm and the State Dependent Riccati Equation (SDRE) technique. Unlike the ordinary methods which deal with the linearized system, this technique will estimate the unmeasured states of the nonlinear system directly, and this will make the proposed technique effective for wide range of operating points. Numerical example is given to illustrate the effectiveness of the proposed technique.